vit-base_rvl_cdip-N1K_aAURC_256
This model is a fine-tuned version of jordyvl/vit-base_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4424
- Accuracy: 0.8932
- Brier Loss: 0.1751
- Nll: 1.0235
- F1 Micro: 0.8932
- F1 Macro: 0.8934
- Ece: 0.0697
- Aurc: 0.0181
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 256
- eval_batch_size: 256
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Brier Loss | Nll | F1 Micro | F1 Macro | Ece | Aurc |
|---|---|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 63 | 0.3624 | 0.8972 | 0.1554 | 1.1865 | 0.8972 | 0.8975 | 0.0426 | 0.0165 |
| No log | 2.0 | 126 | 0.3717 | 0.8965 | 0.1582 | 1.1519 | 0.8965 | 0.8967 | 0.0460 | 0.0170 |
| No log | 3.0 | 189 | 0.3988 | 0.8922 | 0.1690 | 1.1118 | 0.8922 | 0.8927 | 0.0578 | 0.0177 |
| No log | 4.0 | 252 | 0.4032 | 0.8942 | 0.1677 | 1.0854 | 0.8942 | 0.8946 | 0.0590 | 0.0177 |
| No log | 5.0 | 315 | 0.4195 | 0.894 | 0.1706 | 1.0664 | 0.894 | 0.8942 | 0.0628 | 0.0179 |
| No log | 6.0 | 378 | 0.4251 | 0.8955 | 0.1711 | 1.0462 | 0.8955 | 0.8957 | 0.0637 | 0.0179 |
| No log | 7.0 | 441 | 0.4341 | 0.8925 | 0.1726 | 1.0210 | 0.8925 | 0.8927 | 0.0682 | 0.0181 |
| 0.057 | 8.0 | 504 | 0.4379 | 0.893 | 0.1744 | 1.0253 | 0.893 | 0.8932 | 0.0687 | 0.0180 |
| 0.057 | 9.0 | 567 | 0.4411 | 0.8928 | 0.1748 | 1.0200 | 0.8928 | 0.8929 | 0.0712 | 0.0181 |
| 0.057 | 10.0 | 630 | 0.4424 | 0.8932 | 0.1751 | 1.0235 | 0.8932 | 0.8934 | 0.0697 | 0.0181 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.2.0.dev20231002
- Datasets 2.7.1
- Tokenizers 0.13.3